start project

tests/test_real_log_fixtures.py

@@ -0,0 +1,37 @@
+"""Regression test against real production log samples (Chapter 12): both
+files were provided during development after a real cPanel deployment and
+confirmed to parse cleanly with zero failures against the default
+Combined-format parser. Locked in here so a future change can't silently
+break compatibility with this real-world data.
+"""
+from __future__ import annotations
+
+from pathlib import Path
+
+from app.services.log_parser.apache import combined_parser
+
+FIXTURES_DIR = Path(__file__).parent / "fixtures"
+
+
+def _count_parse_failures(path: Path) -> tuple[int, int]:
+    parser = combined_parser()
+    total = 0
+    failed = 0
+    with path.open(encoding="utf-8", errors="replace") as f:
+        for line in f:
+            total += 1
+            if parser.parse_line(line) is None:
+                failed += 1
+    return total, failed
+
+
+def test_ssl_log_fixture_parses_with_zero_failures():
+    total, failed = _count_parse_failures(FIXTURES_DIR / "hemmatitasks.ir-ssl_log")
+    assert total == 222
+    assert failed == 0
+
+
+def test_plain_log_fixture_parses_with_zero_failures():
+    total, failed = _count_parse_failures(FIXTURES_DIR / "hemmatitasks.ir")
+    assert total == 66
+    assert failed == 0